English

Scaling nnU-Net for CBCT Segmentation

Computer Vision and Pattern Recognition 2024-12-03 v2

Abstract

This paper presents our approach to scaling the nnU-Net framework for multi-structure segmentation on Cone Beam Computed Tomography (CBCT) images, specifically in the scope of the ToothFairy2 Challenge. We leveraged the nnU-Net ResEnc L model, introducing key modifications to patch size, network topology, and data augmentation strategies to address the unique challenges of dental CBCT imaging. Our method achieved a mean Dice coefficient of 0.9253 and HD95 of 18.472 on the test set, securing a mean rank of 4.6 and with it the first place in the ToothFairy2 challenge. The source code is publicly available, encouraging further research and development in the field.

Cite

@article{arxiv.2411.17213,
  title  = {Scaling nnU-Net for CBCT Segmentation},
  author = {Fabian Isensee and Yannick Kirchhoff and Lars Kraemer and Maximilian Rokuss and Constantin Ulrich and Klaus H. Maier-Hein},
  journal= {arXiv preprint arXiv:2411.17213},
  year   = {2024}
}

Comments

Fabian Isensee and Yannick Kirchhoff contributed equally

R2 v1 2026-06-28T20:12:49.236Z